I have Nethserver running as a virtual machine (on Proxmox). I have 4 Promxox hosts running all my VM’s. I will be plugging in (via USB) to each Proxmox their assigned UPS battery backup. How would I best set this up from Nethserver. Does Nethserver become the master and each Proxmox Host becomes the slave?
I would like to explore if my Nethserver can manage power outages for all my Proxmox Servers and shut them down if required…and if necessary shut down it’s own host server (but last after issuing the command to shutdown the other Host servers).
Each Host has it’s own Cyberpower UPS and they are all in the same room and same lan (subnet). I am looking to install NUT on each Proxmox host and was thinking of slave but I think you are suggesting I install on each Node NUT as master and let them each handle the shutting down of the VM’s and the host if my UPS kicks in.
The above scenario works…I guess I was looking for a central place I could look and see the status of each UPS for each host which is why I inquired about what the Nethserver UPS module can do for me.
I’m running NUT directly on my Clients ProxMox, in some cases I’m using a Raspberry as NUT-Master. Both concepts work VERY well.
Originally I thought of either using a Synology NAS (Has NUT built in) or NethServer (Also with NUT) as NUT-Master. Alas, both have Users Hardcoded - and not compatible with each other.
So in came Raspberry (Or - in some cases ProxMox directly). I use the Raspberry when I have more than one ProxMox under the same (Powerful) UPS. If I were to use Proxmox in this cluster environment, and THAT Proxmox goes out of order…
I wanted something easily replaceable, and not dependent on a single ProxMox server - to allow ProxMox Cluster functions provide HA like intended.
Using a full NUT stack allows you to configure users / network stuff as needed.
Meaning you can select one or more NUTs to server NUTcgi, displaying all UPS Hosts and their status. Or add in Zabbix Monitoring with UPS Monitoring and all else to your Network…
But if all ProxMox have their own UPSes, by all means, go that way - I would too!
Proxmox shuts down all VMs, and itself when the UPS indicates so.
Another note about UPS:
APC is really moving fast away from OpenSource…
I have some using Reillo - they have VERY goog driver support for NUT…
Not really an option for you with the units you have, but for future readers, a networked UPS greatly simplifies such matters. That way, each slave can talk to the UPS directly–just enter the IP address as the “port” (in the case of nut) and off you go.
In this particular case, though, it isn’t clear that the guest VMs need to talk to the UPS at all. Have the host talk to the UPS, and when the time comes, shutdown the guests (and the QEMU guest agent is supposed to make that process smoother).